Will We Still Learn Languages When Machines Can Translate? (with Abby Walker)
Live AI translation tools are making it easier to converse across languages without a human translator, a phrasebook, or years of study. So will we all just stop taking Spanish or Mandarin classes? Sociolinguist Abby Walker doesn’t think so. She shares the many reasons we learn new languages, some of which have little to do with functional communication at all. We also hear how one AI-assisted conversation brought two future in-laws together, why machine translation is still so darn buggy, and why some languages are attracting a resurgence of new learners even in the age of translation apps.
Life, Automated is a production of the Ryan Institute on Complexity, at Northwestern’s Kellogg School of Management. The host is Jess Love, senior director. Distributed by KQED.
Episode Transcript
This is a computer-generated transcript. While our team has reviewed it, there may be errors.
Jess Love [Voiceover]: This is Life, Automated, the show where we explore what it means to live, work, and make decisions in a world increasingly shaped by machines. I think it’s fair to say that over the past five years, we’ve been bombarded by all the new things AI can suddenly do. It’s solving these long unsolved math theorems, generating songs that rack up millions of listens, and even colluding with other AIs to break into servers.
There’s so much crazy stuff happening, that a lot of other advances that would’ve seemed mind-blowing to us 10, 15 years ago feel almost pedestrian in comparison. I’d put real-time language translation down as one of these advances. Chat bots and translation apps now draw from a range of different kinds of AI to offer live translation for tons of different languages.
You can hear something in a language you don’t know, like Korean. You can use your phone to translate that speech into a language you do know, like English. “Which way is the restroom?” Offer your response in English. “Oh, yeah, it’s down this hallway here, second door on the right side.” And then it will speak your response aloud in Korean.
You can do this today. People already are doing this. What’s more, a lot of tech companies are developing wearables, think smart glasses, smart earbuds, to make live translation more frictionless. No more fumbling for your phone. This is literally straight out of science fiction. Star Trek’s universal translator, Dr.Who’s TARDIS translation matrix, and from The Hitchhiker’s Guide to the Galaxy, the Babel Fish. For me, this raises some big questions. What do these live translation tools really let us do that we couldn’t do before? Will they ever fully replace human translators? And should my eight-year-old still be taking Spanish classes in school?
Today’s guest is Abby Walker, a sociolinguist at Virginia Tech, where she’s co-director of the Speech Lab. She’s also a longtime friend from my grad school days. I’m Jess Love, and this is Life, Automated from Kellogg’s Ryan Institute on Complexity, distributed by KQED.
Jess Love: Welcome Abby Walker to Life Automated.
Abby Walker: Thank you for having me, Jess. I’m so excited to be here.
Jess Love: So there’s a clip I want to play for you.
Jess Love [Voiceover]: I wanted Abby and I to listen to some tape we recorded from a person who recently used one of these live translation tools. This person’s name is Larry Berliner. Larry’s son was in love, and this was Larry’s first time meeting the girlfriend’s parents.
Here’s the catch. Larry, father of the boyfriend, spoke English but not Chinese. And Kevin, father of the girlfriend, spoke Chinese but not English. So they used a ChatGPT app on one of their cell phones to translate between them. In this first clip, Larry’s just remarking about what this tool did for the two of them.
[Audio of Larry Berliner]: If not for this, I wouldn’t have got to know Kevin at all. I was able to connect with Kevin, meet Kevin, talk to Kevin seriously about myself, about my business, about my thoughts of his own daughter, what his daughter has transformed in my son. So much, right? There was so much I was able to say to him.
Jess Love [Voiceover]: And here’s another part of that conversation, this time with Larry talking about how having the tool helped everyone else in the room, too.
[Audio of Larry Berliner]: We have all been there, where there’s the individual who’s the translator amongst three people or among five people, sometimes eight. And two things happen. One is that translator has to work so hard. You can see how much energy they’re having to invest in translating for everybody and make the room successful. The other thing you see is how slow everything goes. The point is, that’s not what happened in our room because Larry and Kevin were involved in this AI-enhanced live translation mode, and the other four were freed. So they were able to have a free, wonderful, free-ranging conversation, and they were able to have that ability to connect in their way.
Abby Walker: Yeah, I love that example and how it’s not just about the personal interaction, but actually about the group dynamics, too. It shows you the work that’s involved in having a human translator, that someone has to mediate that process for you, and that’s actually work on them.
I hadn’t really thought about that aspect of it, which is kinda cool. If you have a human interpreter, there might be things that you want to talk about that you don’t want them to hear about. And I think this is a common situation a little bit with, we know sometimes when immigrants come to a country where they don’t speak the kind of dominant language, that often their kids become interpreters, as they’re dealing with banks and as they’re dealing with medical providers. But, like if you think about, do you want your daughters to know all of your personal information that you might talk about with a doctor? No, right? And so just getting to choose who hears what information, yeah, that’s going to be really empowering.
Jess Love: Yeah, I mean, it seems like in this example, depending on who is going to have to play that role of translator, is he going to feel comfortable, showering all this attention on perhaps his future daughter-in-law, for instance? If she’s the one translating.
Abby Walker: Yes. Yeah. That would be really embarrassing. Though it may be a cool, slick move as well. She has to be like, “Oh, Daddy’s saying I’m so pretty,” and yeah, something like that.
Jess Love: So already these tools are letting people connect with other people in new ways, really connect. But I didn’t invite Abby on this show just to marvel over what’s now possible with machine translation. Like I said, I have a lot of questions. Like, “Is that it? Are we humans all just done learning second, third, fourth languages because the machines can translate for us now? Or are there some parts of a conversation that a machine probably isn’t going to be able to translate for us, at least not anytime soon?”
So a lot of people think of translation as just this simple process of taking words from one language and then mapping them one-to-one onto another language. But from a sociolinguist perspective, what else is language doing other than delivering that kind of information?
Abby Walker: It’s doing so much, Jess. I love all the other things that language does. So one of the big things that I study and my field looks at is all the other information in the speech signal. So not just the content of the message but how you’re delivering it and all the other information that comes from that. So for example, anyone listening right now, you can probably hear that I was not born and raised in America.
I’ve got a New Zealand English accent. But people in New Zealand can probably also hear that I’ve lived in America for a while, that’s probably in my speech as well. You might be able to hear that you and I are friends and friendly. We’re both trying to be friendly. We’re both trying to sound smart right now, right?
Maybe you can tell that this is a little bit more awkward formality for us than usual conversation with wine. So just your identity, and sometimes that stuff you’re trying to mark, like sometimes you’re like, “I want to sound really professional in this moment. I want to sound really from this place. I want to really lean into my higher roots or my New Zealand roots or something.” But then other times, there’s information there that you just can’t even control, right? Like, I’m not always necessarily trying to be like, “I’m a woman,” but most people will hear my voice and be like, “You’re a woman.” Or I’m not always trying to say I’m from New Zealand, but my vowels are always saying I’m from New Zealand.
Jess Love: Right. So whether you want to or not, you are communicating a whole lot about who you are as a person and maybe how you’re feeling in that moment.
Abby Walker: Yeah.
Jess Love [Voiceover]: Abby says that any tool focused primarily on the words we are saying will always miss a lot of the social information that language communicates, information that a human interpreter who can pick up on body language or emotion could probably make better sense of.
Jess Love: And even at the word level there’s a lot of ways for a translation to be incomplete.
Abby Walker: I mean, some things like if I say boot or trunk of a car, that’s a difference between British and American English. But I’m thinking even things more like if I say groovy or neato or amazeballs or, we have lots of ways to express that something’s cool, and how another language does that, they might miss that I’m being playful, or they might miss that I’m outdated in my slang.
Jess Love [Voiceover]: How do you translate that someone is saying amazeballs unironically because they’re really that excited? How do you translate that someone is saying groovy ironically because they’re making fun of themselves for being five decades late to the party? And that’s not all.
Abby Walker: The other thing that language has built into it, both in vocal tone and stuff, but then also in word choice and sometimes actually in your grammar, is your relationship to other people. It’s not just me expressing my identity. I can also be trying to say something about the relationship that you and I have or the relationship I’d like us to have.
Jess Love: So anything that is doing this translation, whether it’s a human or whether it’s a machine, they’re having to make some choices, and the nature of those choices are not going to be obvious to somebody who’s participating and only able to understand one side of that conversation.
Abby Walker: Yeah, and most of the time, again, this… The technology’s kind of amazing. Being able to talk with anyone who you couldn’t talk to before, if the alternative is not being able to talk to them at all, that’s pretty amazing. But there are going to be times where if you don’t know the situation, it might be doing some really impolite things, and you might not know, right? Like the translation system might have decided to use an informal tone here, and you don’t realize it.
Jess Love: And to some extent, that’s going to be true of a human translator but to a much lesser extent if they are very culturally fluent with both languages.
Abby Walker: Yeah, no, I actually… Now I’m remembering I had a student who was Korean-American, and her parents had come over from Korea beforehand, so they were native Korean speakers, and I was learning Korean at the time. And it was graduation, and I was saying something to her dad, like I don’t know. I can’t remember what I was saying, but I was trying to say something polite. And she’s like, “Um,” kind of to her dad like, “Uh, she’s just learning.” And I was like… I assumed that I’d used some impolite grammatical forms or something, right? I’d done some… She knew that I was kind of doing something inappropriate that I didn’t know that I was doing and was just smoothing over the situation because yeah, I think one of the things that humans do is that we’re very socially savvy. We’re always looking at, contextually, what’s contextually appropriate language. And if you’re a bilingual person, which a human translator has to be, right, to some degree to be able to do this translation work, presumably they’re going to know the social navigation pathways of both languages. And so they’re going to be able to pull on that information.
Jess Love: Is there a danger that if we do move toward this world of relatively frictionless live translation, we could become overconfident that we’ve understood everything?
Abby Walker: Yeah, that’s… I think that’s almost part of it. It’s almost like the same problem that people are having right now with, you know, you hand in an essay that AI produced, and it looks so good that you don’t know that the citations are made-up. It’s that sort of confidence that it got the answer right that you would have that same kind of problem with translation, too.
Jess Love [Voiceover]: Given the sheer complexity of language and all the things it’s trying to do, maybe it shouldn’t be a big surprise that as incredible as these tools are, they still make mistakes. A recent survey found that about half of people who’ve used some kind of translation app, chat bot, or wearable find it clunky.
My own personal experience is pretty hit or miss. When I visited Quebec City this summer, Airbnb automatically translated my text messages to my host into French, and then translated his responses back into English. It was so seamless that I barely noticed. But then, when I tried out a pair of smart glasses to have a spoken conversation with a Bulgarian scholar here at the Ryan Institute about his summer vacation, a topic slightly more nuanced than what is the entry code to the apartment, it was a little rough.
At first, I had to repeat my question three times until the glasses finally heard me well enough to translate.
[Jess Love talking to Bulgarian Scholar]: “All right, let’s try again. Okay. Okay. I was holding it too long. I heard a rumor that you are going to the Dolomites this summer to do some hiking.”
[Bulgarian Scholar Translation Audio]:That’s exactly how we’re going with a large group of friends. It’s about 20 people, and I’m pretty excited since it’s raining here, but there are almost no mountains, and I very rarely get to see mountains.
Hungarian Scholar: It’s somewhat accurate.
Jess Love [Voiceover]: The glasses kept messing up. We managed to stumble through a basic conversation but only because my colleague knew English and could correct the mistakes. It was not frictionless. So at least with the tools we have today, it’s pretty clear to me that some tasks will still require human translators, even if a lot of others don’t, and Abby agreed.
Abby Walker: I think it’s the case that we’re always going to need human translators, because there are high-stakes situations, right? Like international negotiations, big business deals. But I also imagine a young person looking at a career path and still seeing, probably rightly so, that there’s going to be less path forward for them there. And I think about the medical world a bit. I think that’s often where there was a feeling that there would be a lot of interpreters was in the medical field.
And you do already see them starting to use these tools and these translation tools automatically. You would hope they would always have someone on staff still, like to do translations when things get really nuanced and stuff. But I do think if I was a young person looking, even though there still will be those jobs, I would wonder about them.
Jess Love [Voiceover]: I would wonder too. Some of the practical reasons for learning a language are going away. After all, not everyone is going to be involved in negotiating corporate mergers or making high-stakes medical decisions. For a lot of situations, good enough really is good enough. And AI’s getting better. Companies are already trying to deal with these issues by like collecting visual data or following people over time so the machines have access to previous conversations.
There’s no guarantee that these tools will still be making the same kinds of mistakes in five years, or even two. And so, yeah, some people who would’ve learned another language probably will decide, “Eh, why bother?” But Abby doesn’t see a future where people just decide to stop learning other languages, no matter how good the machines get, because there are a bunch of reasons for learning a language that have nothing to do with the mere exchange of information.
After the break, she makes her case. We’ll also hear why some very unexpected languages are actually experiencing a resurgence of new learners, apps be damned. Stay with us.
Abby Walker: I’m going to Germany later this year. The big reason, one of the big reasons I wanted to go to Germany, apart from working with some great researchers, is that I wanted to get my German better. I have no need to get German better. This is not an important thing in my life. I will not speak it. I will never need it, but I want to be good at it, right? I want to be better at it. I like it. I enjoy it. It’s a puzzle. It’s a game.
Jess Love [Voiceover]: Abby’s not alone. A lot of people are learning languages they don’t have to learn for purely communicative reasons, like because they need a job or need to get around a city.
Abby Walker: Like people right now are learning Welsh, and people are learning Te Reo Māori, and these people can all speak English in Wales and in New Zealand, but they’re choosing to learn this language that they do not need for communication, so why are they choosing to learn this language? Because they have an identity connection to this language, right? And so its important to them, separate from how many other people they can talk to in this language, right? Like its kind of a way to connect to an identity, because I think people really do associate languages with identity.
Jess Love: Yeah, that’s so interesting. Is there a resurgence in people trying to learn these minority languages, even right now in 2026?
Abby Walker: Yes. There’s definitely a resurgence in people trying to learn these languages. But that’s been happening since the ’80s, ’90s, 2000s. But since I’ve been in the States for almost 20 years now, and when I go back to New Zealand, it is shocking to me the amount of Te Reo that is around now.
Jess Love: And what is Te Reo [Māori]?
Abby Walker: It’s the language. Sorry. It’s Māori. Māori. Sorry. It’s the Māori language. It’s just called the language, so it’s Te Reo.
Jess Love: Yeah, can you just give a little… ‘Cause most of our audience is not from New Zealand, so just give a little… What is this language, and why do you think that people are learning to speak it more in New Zealand?
Abby Walker: Yeah, so this was the first language, human language spoken in New Zealand by the Māori. And then waves of English settlers came, and the language was forbidden, like happens in many of these situations. There was both an incentive to learn English for economic reasons, but then there were also actual moves to ban the language in school. And so kids would get smacked and things if they were speaking it in playgrounds and stuff. And so there was a generation or two that kind of were native speakers of Māori that literally intentionally chose not to pass it on to their children because it had only caused grief for them, right?
And then there was this next generation that was “Wait, we’ve lost this kind of connection to our identity.” And so there’s been this really big resurgence. And this is happening before the technology that we’re talking about, but it’s happening because this has identity meaning to people. This is culturally significant to them. Quite correctly, they see it as something that was taken away from them by kind of outsiders, and then they are trying to correct that and connect to these roots. Which what’s really interesting in New Zealand is a ton of white people or Pākehā, like me, a ton of white people have been learning Māori as well, right? So it’s also becoming a symbol of your political and New Zealand identity as well.
Jess Love: So because this language has more cultural significance than it does, a purely functional significance, do you see it as pretty protected or insulated from some of the challenges of machine translation?
Abby Walker: Yeah. When I’m thinking about it, now that we’re talking about it, I do think that it would be unlikely that you would use a translation app. If you were talking to someone who was a native or near-native fluent speaker of Te Reo and you wanted to communicate with them, you would do it in English or you would learn Te Reo. Why would you use an app to talk to them in this language, stiltingly with pauses and slowly when you both can talk in English, right? They are not going to appreciate that. They’re not really going to be interested in that, and what would probably be more impressive was you stumbling through some beginner Te Reo, which at least showed that you were putting in some effort yourself, right?
Abby Walker: Do you remember that Mamdani clip? Should I try to find it and play it?
Jess Love: Okay.
Jess Love [Voiceover]: Abby wanted us to listen to this recent video released by Zohran Mamdani, the mayor of New York City.
Abby Walker: So there’s this clip you can find, it went kind of a little viral, a little while ago, and it’s Mamdani doing… the mayor of New York doing a piece to camera, and it’s kind of some political speech. I don’t know what he’s actually promoting.
[Audio of Zohran Mamdani speaking in Spanish]
And so he is talking in Spanish and so he is trying to, you know, reach his Spanish Constituence. He does not speak Spanish, right? And so, there is this video of him and he is kind of like stumbling through Spanish and he takes it again and he’s getting corrected by AOC on his pronunciation. He keeps kind of forgetting things. And they’ve released this, right? And they think it’s not crazy that that happened but it’s very interesting to me that they released it because they are trying to emphasize that it wasn’t easy for him to speak Spanish and that he needed help speaking Spanish and he is not good at Spanish, but they knew that it would be charming and it is, right? Because it shows someone is trying to speak Spanish, right?
[Audio of Zohran Mamdani speaking in Spanish]
They could have just released speeches with him in like fluent Spanish, or they could have had someone else who is a fluent Spanish speaker speaking on his behalf. In a little while, they could probably do videos where it looks like he is speaking Spanish, right? We’re kind of at that point of the technology now with video and audio that we could have something that sounds like him speaking in Spanish, looks like him speaking in Spanish. But they don’t do that, right? What they do is show someone who isn’t very good and who is trying, right? Part of what is really important to people is that you try to speak their language, right? And an app takes out that friction. It makes communication easier, but part of what language is signaling, too, when you use a language that is not a language that comes easy to you, is you’re signaling that you are trying, right? You are signaling that you are putting in some effort, that you don’t expect them to just understand you, and that you’re putting in some effort to be understood.
Jess Love: All right. Well, we’ve been dancing around this idea of the importance of friction, and I want to just ask you to spell it out. Why is friction important? What could friction signal? We’ve talked about belonging, we’ve talked about effort. What else?
Abby Walker: I mean, if I’m really just using an app to talk to you, right, I… And I’m not paying attention, I’m not really trying to do anything, I’m just trying to communicate with you, so I use an app, and it spits out whatever to talk to you, and I nod. You got the message, I got the message, I move on. Like, I haven’t really changed because of that interaction, right? So when you learn a language, to talk to someone, the effort of doing it, you are actually also changing yourself. So you are adding vocab, adding things that you can use in a new situation, adding things that you can use when you don’t have your phone around you, et cetera. So there’s just, literally you are changing a little bit, maybe in small ways. There is some evidence for regular bilinguals, like people who are regularly using multiple languages in their day-to-day, that they have better executive function than monolingual speakers.
And by executive function, we’re talking about your ability to ignore irrelevant information and suppress responses that you might automatically make. And if you think about it, this makes a lot of sense why this would be tied to bilingualism, because you have a day where you are telling whatever language you’re not using, you’re saying, “Shut up. This isn’t about you.” So you’re ignoring anything else that’s irrelevant while you’re having these conversations. And from my understanding, those advantages get bigger as you age. So it’s essentially like you’re keeping your brain more active in this particular way. So obviously, all of those benefits, and just the fact that each time I stumble through something in German or Korean or today Māori, I am getting a little bit better, right? I’m learning a little bit more. That is this really big obvious thing that is being lost, is just that you are not having this growth. And it’s almost sad in some ways, because you had an opportunity, right? So let’s say you go to another country, you stay there for two months. You have this opportunity to do something, and you just spend all the time using these translation apps. You miss this opportunity to have this personal growth.
Jess Love: It’s true. I went to Italy, and by the end of the trip, I was fully capable of asking where the bathroom was all by myself.
Abby Walker: In Italian?
Jess Love: In Italian.
Abby Walker: Wow.
Jess Love: Don’t ask me to do it now.
Abby Walker: I was gonna say.
Jess Love: Oh, boy. Try. Al bagno? Bagno? Bagno? I don’t want to talk about it.
Abby Walker: Okay.
Jess Love [Voiceover]: Live translation tools are decent, and they’re going to get better. You can even imagine a version that works out most of the bugs. AI that does start picking up on body language and slang. Science fiction becoming reality. The universal translator, the Babel fish. But do we always want that? Throughout our conversation, as we talked about the importance of friction, it struck me that maybe today’s live translation tools, imperfect as they are, have their own charms, not despite the friction, but because of it.
I decided to play Abby another excerpt from our conversation with proud dad, Larry Berliner. While he and fellow dad Kevin were conversing on their ChatGPT app, they were speaking into the phone, which would generate a written translation, which the other one would read. Hardly seamless. But here, Larry talks about the conversation being so profound because of those seams.
[Audio of Larry Berliner]: This allowed us to speak in long paragraphs, like I’m doing right now, right? Shakespearean style, not a back and forth, but long paragraphs of ideas, and then it would translate, and we’d watch each other read it, which was exciting, and then the person would respond. And what ended up being magical was, first of all, we did it for a very long time.
What he chose to talk about was things that I don’t think he would ever have spoke about first if I was speaking his native language. Or wow, the extent of which is amazing. He talked about how he had met Sonya and the previous people he had been with at the time. He talked about his failures as a father. And just– it was immediately right to the bone.
Jess Love: I’m just struck by how he feels that conversation was facilitated by the fact that there was this level of friction, of literally having to read. Maybe the fact that you’re putting something into print and then somebody is going to read it makes you less likely to just waste that time to say, “Well, what do you think of the weather?”
Abby Walker: Yeah, I mean, that’s a really interesting idea, and it kinda makes sense. What it makes me think, though, is that there was this feeling that this was temporary, right? That this was not going to be the regular state of affairs. It’s almost like deathbed conversation. It feels like a little bit like what it sounded like to me. Just “I need you to, here are some really big things on my mind and my heart that I just really want you to know now that we’ve got this opportunity.” It is interesting, though, to think about, what if you never get there with someone who speaks your own language because you’ve always got the dance of we can speak so easily to each other.
Jess Love: Yeah, and it was the artificiality of their own communication that led to something much deeper.
Abby Walker: I love that.
Jess Love [Voiceover]: So am I going to encourage my eight-year-old to learn Spanish? Yes, absolutely. There’s so much we can’t know about AI and where it’s headed, but even if some reasons for learning another language go away, others won’t, or could even grow in importance. And in honor of our topic this week, we’ve machine translated our credits into ‘Klingon and back’, because why not?
Here we go.
[Klingon words] I am Jess of the clan Love. [Klingon words] At the Northwestern University Citadel of Knowledge, inside the Kellogg School of Management, Life, Automated is an honorable battle quest of the Ryan Institute on Complexity. KQED hurls our battle transmissions across the void to conquer the airwaves.
Jesse Dukes sliced the captured sounds into submission. Stephen Jackson forged the battle hymns and struck the war drums. The machinery masters, Will Feeney and George Christensen, trapped the vocal cries of the interrogation inside the sound cage. Ananya Mallapragada commanded the proclamations of our glory and subjugated the civilian chatter web.
Producer: Jesse Dukes
Music: Steven Jackson
Recording: George Christensen, Will Feeney
Promotional help from Ananya Mallapragada
Editorial guidance from Laura Pavin
Special thanks to Larry Berliner and Georgi Demirev for their help testing out live translation tools